数据集：基于PaddleX的人脸、戴口罩和吸烟目标检测.zip

# 基于PaddleX的人脸、戴口罩和吸烟目标检测 ## 一、安装PaddleX ```python !pip install paddlex==2.0.0 ``` ## 二、解压数据集 ```python !unzip -oq /home/aistudio/data/data128757/VOC_MASK_smoke.zip ``` ## 三、生成数据集文件 PaddleX支持VOC格式数据， ## **安装Paddlex** 首先安装1.83以上版本的Paddlex（1.83一下版本的已经不支持此划分命令） ```python #如果下一句代码报错，请先运行此代码 #!pip install paddlex==2.0rc -q ``` ```python !paddlex --split_dataset --format VOC --dataset_dir VOC_MASK_smoke --val_value 0.2 --test_value 0.1 ``` ## 四、数据预处理 这里使用了图像混合、随机像素变换、随机膨胀、随即裁剪、随机水平翻转等数据增强方法。 ```python from paddlex import transforms as T train_transforms = T.Compose([ T.MixupImage(mixup_epoch=250), T.RandomDistort(), T.RandomExpand(im_padding_value=[123.675, 116.28, 103.53]), T.RandomCrop(), T.RandomHorizontalFlip(), T.BatchRandomResize( target_sizes=[320, 352, 384, 416, 448, 480, 512, 544, 576, 608], interp='RANDOM'), T.Normalize( mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225]) ]) eval_transforms = T.Compose([ T.Resize( 608, interp='CUBIC'), T.Normalize( mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225]) ]) ``` ```python train_dataset = pdx.datasets.VOCDetection( data_dir='VOC_MASK_smoke', file_list='VOC_MASK_smoke/train_list.txt', label_list='VOC_MASK_smoke/labels.txt', transforms=train_transforms, shuffle=True) eval_dataset = pdx.datasets.VOCDetection( data_dir='VOC_MASK_smoke', file_list='VOC_MASK_smoke/val_list.txt', label_list='VOC_MASK_smoke/labels.txt', transforms=eval_transforms) ``` ## 五、训练模型 这里定义了一个YOLOv3，使用DarkNet53作为主干网络（该参数训练50epoch差不多就收敛了）； ```python import paddlex as pdx num_classes = len(train_dataset.labels) model = pdx.det.YOLOv3(num_classes=num_classes, backbone='DarkNet53') model.train( num_epochs=270, train_dataset=train_dataset, train_batch_size=8, eval_dataset=eval_dataset, learning_rate=0.000125, lr_decay_epochs=[216, 243], warmup_steps=1000, warmup_start_lr=0.0, save_interval_epochs=5, save_dir='output/yolov3_darknet53', use_vdl=True) ``` ## 六、模型检测结果及可视化 ```python import paddlex as pdx eval_details_file = 'output/yolov3_darknet53/best_model/eval_details.json' pdx.det.draw_pr_curve(eval_details_file, save_dir='./output/yolov3_darknet53') ``` # 目标检测/实例分割准确率-召回率可视化  ```python #预测模型 import paddlex as pdx model = pdx.load_model('output/yolov3_darknet53/best_model') image_name = 'VOC_MASK_smoke/JPEGImages/1_Handshaking_Handshaking_1_164.jpg' result = model.predict(image_name) pdx.det.visualize(image_name, result, threshold=0.5, save_dir='./output/yolov3_darknet53') ``` 2022-02-20 15:29:23 [INFO] Model[YOLOv3] loaded. 2022-02-20 15:29:23 [INFO] The visualized result is saved at ./output/yolov3_darknet53/visualize_1_Handshaking_Handshaking_1_164.jpg